Posted by: Josh Lehner | July 27, 2016

Tourism and Oregon’s Economy

Earlier this month Mark and I had the opportunity to discuss tourism and its impact on the Oregon economy to the Transient Lodging Tax Workgroup, setup following the passage of HB 4146 which raised the statewide lodging tax. Our slides are below. However first I wanted to provide a few summarizing thoughts.

Travel and tourism has been booming in recent years. Nationally the share of consumer spending spent on travel and tourism really has never been higher. Hotel occupancy across the country is at or near record highs. Hotels also have more pricing power today given the strong demand even as new hotel construction picks up.

In terms of employment — borrowing the Federal Reserve Bank of Kansas City’s methodology — the sector in Oregon has just recently regained all of its lost jobs during and after the Great Recession. Consumer spending is back and now too are the jobs across the state. It is no secret that most travel and tourism jobs are low-wage and the industry overall has lower productivity, or value-added in state GDP calculations, than nearly all other sectors in the economy. Even in the economic multiplier world, leisure and hospitality sectors are in the 1.3 or 1.4 range. However there are a few related points here worth mentioning.

First, the flipside of low productivity and low wages coupled with strong industry demand is that it is a large employment sector. 1 out of every 8 jobs in Oregon are within a sector directly tied to travel and tourism. Now, not every single job in these sectors is dependent upon it, but the sectors in a broad sense do rely upon spending from visitors more than some other parts of the economy. This share of jobs varies across the state and the travel and tourism industry has a wide geographic footprint. 28% of all such jobs in Oregon are outside of the Willamette Valley, matching the shares seen in the overall population and public sector employment. Other industries, like the high-wage and fast-growing Professional and Business Services are more heavily concentrated in the urban centers. So the industry is not only important overall but particularly so in some regions like the North Coast where 1 out of every 4 jobs are travel and tourism related and 1 out of every 4 homes are second homes or vacation homes.

Second, these jobs do fill a void and a need in the economy. This is something we (well, me) did not cover in our initial presentation. Representative Davis asked us about it and for good reason. The point being that in the ideal world the economy would produce employment opportunities for everyone along the skills spectrum (and also geographic spectrum too). Of course this is not always how the world works. Job polarization in recent decades has resulted in strong growth in the high- and low-wage occupations. The strong growth in travel and tourism jobs are a result of the strong demand for these types of services in the economy. They can also not easily be outsourced and certainly cannot be offshored, although automation plays a role here too.

Overall the outlook remains bright for the sector. The Baby Boomers today are in their peak travel and tourism years which coincides with their late career and active retirement years. Based on demographics and consumption patterns, growth will likely be slower in the coming decade than in the past, however it is certainly a positive outlook. The Millennials today do not spend as much on travel and tourism but this is due to their age and income levels. As they move into their peak-career and family years, spending will increase, offsetting the future decline of Boomer spending. What does remain an open question are tastes and preferences of the younger generations. They will travel and spend in the future, however will their preferences match their parents’ or will they prefer different locations and activities in addition to lodging type?

Here are our prepared slides for the work group.




Posted by: Josh Lehner | July 21, 2016

Rural Oregon’s Potential Labor Force

In our office’s Rural Oregon report we mention that much of the discussion focuses on data and trends that are backward looking. They indicate how many jobs were lost in the 1980s or how old the typical resident is and the like. While these statistics help describe the current lay of the land, they do not necessarily tell us what tomorrow may bring. To be sure, many of the more forward-looking indicators are also less bright in much of rural Oregon than in urban Oregon, but not all hope is lost. In fact, if anything, some of pessimism about rural Oregon today may be a bit overdone. The reason? Demographics. Yes, that’s right, demographics.

One aspect our office has been researching is potential economic growth and the impact of the aging demographics. Specifically we looked at rural counties and their potential labor force using the same methodology we did for the statewide demographically-adjusted labor force participation rate. What this does is take a more nuanced and I’d argue better look at what is essentially the working age population in rural areas.

What the analysis shows is that while the Baby Boomers’ retirement will weigh on net growth rates across the state, in much of rural Oregon this slowdown has already taken place. It is in the rear-view mirror. Moving forward, the potential labor force will pick up and in some regions it already has.

Why is this? What matters from an economic sense is when an individual ages from their prime-working and peak-earning years into retirement. Unless another individual takes their place, this represents a decline in the economic potential of the economy, everything else equal. And in much of rural Oregon there has been a decline in the potential labor force given demographic and population trends in recent decades.

However, rural Oregon is nearly all the way through this demographic drag. For example, Southeast Oregon is 80% of the way through their demographic drag. The North Coast is three-quarters of the way through theirs. Southwestern Oregon has seen the largest regional decline in its potential labor force, however this has already occurred. Further declines are not expected for the region overall based on our office’s population and demographic outlook, even as individual county performance varies (more on this below).


Some regions, like the Gorge and Northeast Oregon, experienced a slowdown in their potential labor force but the influx of younger individuals into these regions has been enough to offset the retirements. As such there has not been an overall decline in their potential labor force. This is the pattern seen among urban Oregon as well – a relative slowdown but not outright declines.

However, as mentioned above, individual counties do differ, even neighboring ones. Lets take a quick look at the demographics of Klamath and Lake. Here is the 2015 population distribution by age group. Both counties skew older than the state as a whole, but Lake considerably more than Klamath. In fact the largest age groups in Lake are all between 50 and 69 years old, or the transition period from prime-working and peak-earnings years to retirement. As such, Lake has so far not seen much of a decline in their potential labor force, but are about to in the coming decade or two. While Klamath’s largest age groups also are 50-69 years old, the younger demographic groups are considerably larger in relative size and when combined with the population outlook this results in a smaller potential decline over a shorter time period.PopDist15

The last graph shows the potential labor force declines for each individual rural county in Oregon. Continuing with the Klamath and Lake example, you see that Klamath is about 85% of the way through their demographic drag while Lake is just 9%, due to the different demographic structures and outlooks.


As stated at the top, much of the focus on rural Oregon is about what has already happened in recent decades. There is no question that many of these trends paint a relatively bleak picture. However, when it comes to gauging the potential labor force, even one of the most commonly cited trends in rural Oregon, that of the older demographics, indicates that not all hope is lost. It is certainly possible that the actual number of jobs and the actual size of the labor force will remain smaller in the future than in the past. However demographic trends and the population forecast suggest that the outlook is brighter than the conventional wisdom suggests. Many challenges remain of course, but demographic pessimism is likely overstated.

Posted by: Josh Lehner | July 19, 2016

Oregon Employment Update, June 2016

This morning the Oregon Employment Department released the initial estimates for June 2016 in terms of jobs and the unemployment rate. Overall the data continue to bring good news, even as the unemployment rate ticked up a bit. The reason being is the state continues to add jobs and the labor force is growing. As shown below, our office’s measures we use to gauge labor market slack continue to improve — certainly over the past year.

First, it is estimated that Oregon added 3,000 jobs last month. Overall the job gains have slowed in recent months relative to late 2015 and early 2016. This is a similar pattern as seen in the U.S. data. However those strong gains of 5,000 or more per month are not sustainable over a long period of time. Such gains are only seen during peaks in the business cycle. Most importantly for Oregon’s economy is the fact that the recent gains are still more than enough to keep pace with population growth.


Furthermore, Oregon has now reached another milestone in terms of the recovery and expansion. Not only has Oregon added enough jobs to regain all the Great Recession losses, the state now has caught back up with population growth. Even as the economy cratered, people kept moving to Oregon and the state’s Jobs Gap hit nearly 170,000 or 10% back in early 2010. Today this gap is now closed.


Our office’s outlook has the economy to add jobs at a full-throttle rate until the labor market slack is gone, and for the economy to even run a bit stronger for a little longer than that. So our concern is that if jobs have caught up with population growth, the slowdown in job gains to a more sustainable rate of growth may come earlier than what we have in our outlook. This is something we discuss regularly and is a key factor in the timing of our forecast. So while jobs have caught up with population, the state’s Total Employment Gap still indicates some slack in Oregon, albeit considerably less than in the recent past. Right now the labor market is as tight as it was during the peak of the housing boom. However there remains room for improvement. In fact, the gap increased 0.27 percentage points from May to June. This was due to both the unemployment rate and the share of Oregonians working part-time but want full-time work increasing. These gains were offset somewhat as the labor force participation rate also increased, which is a good thing.


All told the Oregon economy continues to improve. It is expected that job growth will slow to a more sustainable rate as economic slack diminishes. However our office’s forecast calls for full throttle rates of growth through the remainder of the biennium. Should the slowdown occur a bit sooner, then we will adjust our outlook accordingly. To date, however, withholdings out of Oregonian paychecks continue to show robust gains. In fact our growth is among the strongest in the nation based on recent conversations with out counterparts in other states.

Posted by: Josh Lehner | July 7, 2016

Office Support Decline Rivals Manufacturing

When it comes to middle-wage job losses, much of the focus is on the blue collar occupations that are generally held by men without college degrees — construction, installation/maintenance/repair, production, and transportation. The story of how globalization, technological change, and other factors have hollowed out these jobs is well known. Not only are their fewer such job opportunities, but their relative wages have eroded to the point where, say, the remaining manufacturing jobs no longer pay much of a premium compared to the economy overall. This is one reason nonparticipation rates are rising among prime working age men without college degrees.

However, there is an important aspect to the decline of middle-wage jobs that is generally not discussed as much: their impact on women. Whenever I give a presentation on our office’s state level job polarization work, I always talk about Mad Men. In the office scenes of the show, there is always one administrative assistant outside of each office. No modern workplace looks like that. Today, one office support specialist serves multiple managers instead with relative ease. Increased productivity can be seen due to computers, software, travel websites and the like. This speaks nothing about other office support occupations like switchboard operators, file clerks and typists which have effectively been eliminated entirely. My back of the envelope calculations show the decline of office support jobs in the Oregon economy overall from 2000 to 2015 has resulted in nearly 50,000 fewer such jobs — for both sexes and all ages — due to this increased productivity and changing workplace practices. That’s equal to one full year of good job growth in the entire state, so a huge number.

While all of the above is true and I regularly discuss these trends on both blue and white collar jobs, I still was not fully prepared for these results. The decline of office and administrative support jobs for women with a high school diploma or less rivals the decline of production jobs for men with the same educational attainment here in Oregon.


In the 1970s, about 1 in 5 prime working age Oregon men with a high school diploma or less worked in a production job, essentially the manufacturing jobs that actually do the manufacturing. In the 1980s, about 1 in 5 prime working age Oregon women with a high school diploma or less worked in an office and administrative support job. Today, employment rates in these occupations is roughly half of what it was a generation ago for these demographic groups. The production declines for men are more severe, however the decline of office support jobs for women is certainly of a similar scale.

Middle-wage job losses are not just a blue collar male issue; women have seen large erosion in employment opportunities as well. Where the gender trends differ somewhat, at least in Oregon, is that low-wage work among women without college degrees has increased in recent decades, whereas for men the increases in this type of work are pretty minimal.

Additional Notes:

As a check on the data used here, which has a relatively small sample size, I also looked at Census data from 1980 to 2014. It shows an EPOP decline for males in production jobs of 9.4 percentage points and for women in office support jobs of 7.1 percentage points. So roughly similar to the graph above. What does differ in the Census/ACS data is that the decline for women in office support occupations has all been since 2000. This decline has been particularly large from 2007 to 2014.

One thing that is clear in the ASEC data is that in the past 5-10 years, the share of office support jobs going to women with a college degree has risen considerably. From the late 1980s through 2010, the share of female office support workers with a college degree was 15-20% of all such jobs, among prime working age Oregon women. The 2013-15 average is 34%. This is a massive change. It is also confirmed by anecdotal reports from our friends at Employment, like Christian Kaylor, who talks about how businesses today use a college degree requirement for office support jobs as a mechanism to screen the applicant pool, even though the jobs generally do not require a degree. The data confirm this is having an impact on who gets the jobs.

Finally, in the Census data I also looked at the U.S. overall. Here the decline of men in production jobs with a high school diploma or less declined by a similar margin as in Oregon (9.1 percentage points nationally). However women in office support jobs only fell 4.6 percentage points over this time, considerably smaller than in Oregon. I do not have a good reason for why this differs as the historical figures for 1980, 1990 and 2000 are very similar for Oregon and the U.S. One hypothesis could be the increased number of women with college degrees in office support jobs in Oregon has been larger than national trends. Thus the shift of educational attainment within these jobs is driving the larger Oregon decline. However, that is just a hypothesis at this point.

Posted by: Josh Lehner | July 1, 2016

The Elephant Graph (Graph of the Week)

A few months ago former World Bank lead economist, Branko Milanovic, released a new book titled Global Inequality. His work has taken the economics profession by storm since. One chart in particular, dubbed by some as the elephant graph, because, well, it looks like an elephant, tells a fascinating story. Branko has a new, very accessible article out today over on VoxEU. I am using that article and chart for this edition of the Graph of the Week.

What the elephant graph shows is “cumulative real income growth between 1988 and 2008 at various percentiles of the global income distribution.” While that’s a mouthful, these economic and income trends in recent decades are important. As discussed below, the elephant graph is also very relevant to our office’s recent work on prime working age Oregon men and women, including some more to come next week.

Branko labels 3 points (A,B,C) which are of particular importance:

The results show large real income gains made by the people around the global median (point A) and by those who are part of the global top 1% (point C).  It also shows an absence of real income growth for the people around the 80-85th percentile of the global distribution (point B).

Branko goes on to explain which populations are at points A, B and C. “Nine out of ten people around the global median [point A] are from Asian countries, mostly from China and India. These gains are not surprising, given that Chinese and Indian GDP per capita has increased by 5.6 and 2.3 times, respectively, over the period… Such dramatic changes in relative income positions, over a rather short time period, have not occurred since the Industrial Revolution two centuries ago.”

Point C consists of the global top 1%, which is “overwhelmingly people from the advanced economies – one half of the people in that group are Americans…”

Point B has been getting the most attention because it shows, effectively, no income growth in recent decades. And the fact that “Seven out of ten people at that point are from the ‘old rich’ OECD countries. They belong to the lower halves of their countries’ income distributions, for in effect the rich countries’ income distributions start only around the 70th percentile of the global income distribution.”

This gets at the overall stagnating median household income figures seen in the U.S. and here in Oregon too. It also gets at the lower inflation-adjusted wages for prime working age adults without a college degree, job polarization and the like. Again, I will have a bit more on this next week from an Oregon perspective.

What Branko’s elephant graph clearly shows is how these trends compare to the global changes seen in recent decades. In his article today, and in various book reviews you can find online, the discussion broadens to talk about inequality and the impact of globalization and the like, in addition to some political ramifications of these trends. However, for today I just wanted to highlight this new important work and how it helps place some trends we’re seeing in the U.S. and here in Oregon in a global perspective.

Posted by: Josh Lehner | June 30, 2016

Prime Working Age Oregon Women

Following our look at labor force participation rates (LFPR) and employment opportunities for prime working age males in Oregon the other day, today we will focus on female participation. The patterns shown below are both different from a 60 year perspective but similar if we focus just on the past 15 years or so.

Female participation differs significantly in the sense that it rose considerably in the 1960s, 1970s and 1980s as women entered the workforce in much greater numbers than historically had been the case. These gains are seen across the educational attainment spectrum. Female participation, while higher than a generation or two ago, still remains lower than male participation. Nationally, in 2015 prime working age women had a LFPR of nearly 74% while men were just over 88%. According to the Oregon Employment Department’s report on participation, it looks like the male-female gap is somewhat smaller here in Oregon.


Where male and female participation trends look similar is in the fact that nonparticipation among prime working age adults has been rising in the past 15 years. It is certainly true that we are seeing in increase in stay-at-home moms, as our office’s report shows (and looking at the most recent data these trends have continued).

However, given lower birth rates and the like, the uptick in stay-at-home moms is not quite as apparent when looking at all women in terms of why LFPR has been decreasing. Overall, much of the gains in nonparticipation are due to reasons other than staying home to take care of the kids, even as that remains the single largest reason cited. Since 2000, nonparticipation rates are up due to increased school enrollment (1.0 percentage point), retirements/other (1.2 ppt), illness or disability (1.4 ppt), those who could not find a job (1.8 ppt) and taking care of home or family (2.4 ppt).


The changing labor market and job opportunities likely has a significant impact on participation rates for women, just as they do for men as we noted the other day. This is particularly true among those with lower levels of formal schooling. The last graph shows the employment-population ratio by job type for Oregon women with a high school diploma or less. This is the time series version of the distributional change chart included in the male post.


While low-wage employment opportunities have been rising over time, the decline of middle-wage jobs is equally important for women in the workforce. This is true, even though the conventional wisdom tends to focus on the decline of the blue collar jobs which have traditionally been held by men — construction, manufacturing, transportation, etc. In fact, next week I have some follow-up graphs on middle-wage job losses for prime working age Oregon men and women. Not to be click-baity, but one result even shocked me even as I knew it was probably true.

Posted by: Josh Lehner | June 28, 2016

Prime Working Age Oregon Males

The Oregon economy is nearing full employment, a milestone not seen since 2000. According to our office’s Total Employment Gap, the one remaining part that is not fully healed is the participation gap. This measures the difference between the actual labor force participation rate — the share of adults with a job or looking for work — with a demographically-adjusted version to take into account the aging Baby Boomers. The underlying question is how much slack is left in the economy? The reason being is that once the slack is gone, one can expect the economy to begin to transition to a more sustainable rate of growth in the future.

So it was with great interest that the White House Council of Economic Advisers released a new report focusing on participation among prime working age males, those between 25 and 54 years old. One reason this group is analyzed is they have historically had the highest participation and employment rates in the economy. This group is also seeing long-run participation declines for much of the past 50 years. This matters for both macro reasons, such as potential economic growth, and micro reasons:

A large body of evidence has linked joblessness to worse economic prospects in the future, lower overall well-being and happiness, and higher mortality, as well as negative consequences for families and communities.

To help our office better understand these changes, what follows is an abridged version of the CEA report using Oregon data.

First, there have been considerably different outcomes among various subgroups within the prime working age population. One key aspect is educational attainment. Those with a college degree have not seen their employment or participation rates decline nearly as much as those with less formal schooling. Oregon’s patterns look nearly identical to the nation overall.


Next the CEA tries to understand what is driving these trends by analyzing supply side and demand side issues in addition to broader institutional changes. Their results show that while supply issues are real, they alone cannot explain the timing nor the magnitude of the LFPR declines. Demand side issues fit the data better.

On the supply side, participation among prime working age Oregon males is down 6 percentage points since 2000. This is a big drop. There has been a large increase in those claiming they are not looking for work due to being ill or disabled. This increase accounts for 60% of the participation rate change overall, on net. However, what is interesting to note is that this increase has not been accompanied by a corresponding increase in those actually receiving disability payments. There is a big disconnect here that may be a little hard to reconcile. However the data makes clear that the increases in those claiming disability as the reason for not working is many times larger than the actual increase in those in disability programs.


To the extent that these individuals are permanently gone from the economy, it does not necessarily matter if what they claim and what they do are different. A smaller labor force results in slower economic growth moving forward, everything else equal.

What the CEA does find is that demand side issues fit the LFPR declines better overall. This means that the fewer job opportunities for prime working age males, particularly those with less schooling, leads to lower participation rates. This is both good and bad news. A stronger economy that creates more jobs will employ more people. However to the extent that the skill sets of those out of the labor force today do not match the demands of the new jobs, a stronger economy alone cannot help these trends.

Among Oregonians with a high school diploma or less, middle-wage jobs have declined in recent decades. This is true both as a share of the economy and in the outright number of such jobs. Of course we have known about job polarization and its impact on the Oregon economy for some time now. However, the losses of blue collar jobs is particularly pronounced among this demographic group. While some have found work in either high- or low-wage occupations, the vast majority of this adjustment has been to drop out of the labor force entirely.


Part and parcel with these trends is the flat to declining wages for those with a high school diploma or less. Two generations ago, wages were somewhat similar across the educational attainment spectrum, both nationally and in Oregon. This widening gap is due to both wage gains for college-educated workers and also wage declines for those with a high school diploma or less. Additionally, the CEA report found that states with higher wages for those in the lower part of the income distribution saw higher LFPR overall.PAMwages

The falling participation rate is a discouraging economic trend, particularly among the prime working age population. In recent years our office has researched various portions of the LFPR decline, including the increase of stay-at-home parents, and to what extent the declines are more likely to be cyclical vs structural. This last part is the most important for our office and the outlook both in trying to time the transition to more sustainable rates of growth and also the size of the potential labor force and economy in the future.

For businesses, we know that as the economy gets closer to full employment, finding quality workers is certainly more challenging. Businesses must cast a wider net and also pay higher wages to attract and retain the best workers. This is happening today, in Oregon in particular where our LFPR is rising quickly and wage growth outpaces the typical state. Of course there is still room for further improvements too.

The CEA report itself explore more in-depth the impact of spousal income, public sector benefits, poverty rates, international comparisons and flexible labor market policies. The report ends with a section on policies to try and boost participation, for those interested.

Posted by: Josh Lehner | June 23, 2016

Oregon Income, A Discussion

A lot has been made of and discussed about the fact that Oregonians have lower incomes than the nation overall. However, I find myself reaching for various measures of income when the need arises, or to answer specific questions. Are we talking about wages, or about household income, or about the potential tax base for revenues? There is no one measure to rule them all. Each has both pros and cons about its usefulness, depending upon the question being asked. Below I have pulled together a handful of income measures to show how Oregon stacks up. Interestingly enough, the differences shown below are pretty important. Note that the graph shows Oregon’s income by category as a share of the national figure. Values below 100% indicate that Oregon has lower incomes than the nation.


Clearly there is a wide range of outcomes when looking at the data. Median earnings for full-time workers in Oregon are just a notch below the national figure, however the state’s per capita income is 10% below the national figure. There are some important reasons for why these patterns happen, based on the underlying demographics, economy, share of full-time vs part-time workers and the like. However I want to focus on two measures in particular: per capita personal income and median household income.

Per capita personal income is a measure that simply adds up all the personal income in the state and divides by the population. Oregon’s per capita personal income has eroded over time, relative to the nation, as research from both the business community and the Employment Department has documented. (Quick aside: our average wages have not eroded in a similar manner, which is an important distinction) I would argue that per capita personal income is a very important measure when discussing tax revenues, or the potential tax base. It shows the average ability to pay, more or less, of a given region’s residents. However it is also an average and susceptible to both demographic shifts and to the distribution of the incomes across the population. As such, even though per capita personal income is widely cited, it is probably not the best measure to use when talking about how Oregonian incomes compare to the nation overall. I am not saying, however, that per capita personal income is not an important measure, rather that it has a time and place to be used.

Median household income is my go-to measure when someone asks to compare incomes across regions. The simple reason is that, by definition, the median household income represents the typical household, so it is much less susceptible to outliers, distribution issues and household/family/population composition. In my view, it is a more accurate reflection of what the typical Oregonian earns.

Oregon’s median household income has largely tracked the national figure in recent decades, which is to say it is unchanged or declining after adjusting for inflation. Outside of the timber industry’s peak and restructuring in the 1970s and 1980s, Oregon’s median household income has pretty consistently been 2-3 percentage points lower than the national figure.


What is really interesting is to look at the distribution of household incomes in Oregon relative to the nation. The final graph does just this for incomes across the entire distribution, based on percentiles. For example, Oregon households at the 20th percentile earn about $21,000 while a U.S. household at the 20th percentile earns $21,700. As noted in the graphs above, the median Oregon household income is about 5 percent below the national median ($51,100 vs $53,700).


What really stands out is the widening gap at the highest income levels. To be in the Top 5% of household income in Oregon it takes roughly $185,000 per year, whereas to make it nationally it takes roughly $207,000 per year, or a gap of 11 percent. In fact, the Oregon-US gap among the highest 10% of incomes is twice that of the bottom 90% of incomes. Why exactly this is the case is not entirely clear. However we do know that Oregon is not a financial center, so we miss out on some of those really high-paying occupations. It is also true that our high personal income tax rates likely play a role as well in deterring some, but not all, high-income households and businesses from locating or operating in our state. But what is very clear is the fact that the widening gap in incomes at the top end of the spectrum mathematically weighs on average incomes, but not medians. Go back to the first chart and see how much further away from the national figures the averages are relative to the medians. The differences there are being driven by the widening gap at the top.

Is this a problem? Well, that depends upon the question you are asking.

A big thank you to Kanhaiya, our state demographer, for pulling the historical Census data and the distribution figures for the latest ACS!

Posted by: Josh Lehner | June 14, 2016

A Note About The Housing Trilemma

The response and feedback on the Housing Trilemma has been amazing this past week, if not a bit overwhelming at times. A big thank you to those of you sharing, asking questions and providing feedback and thoughts on improvement. I really appreciate it.

Unfortunately in my effort to keep the original post a manageable length, I left out a few additional details that would have helped the public conversation this past week. I want to clarify three things regarding the placement of metros on the Venn diagram, the Midwest/Great Plains, and also the quality of life measures.

First, while only 15 metros were listed on the Venn diagram, every single metro does fit on it somewhere. I only used a few representative examples for each part of the diagram. This has been a point of confusion, due to my failings of explaining that originally.

For example, while I mention that 8 individual metros rank among the top half along all three dimensions, I only show 3 metros in the middle part of the diagram. Additionally, New York slots right in between San Francisco and Portland and Kansas City is just to the right of Oklahoma City in the original work. From a broad regional perspective, here is how the Venn diagram looks. As you can see, these regions and the select metros listed originally line up accordingly.


Second, the Midwest or Great Plains do not generally knock any particular dimension of the trilemma out of the park. Rather, they score very solidly along each dimension. That is why they are placed in the middle, or overlapping region. This does not necessarily mean they are the best place to live in the country — that is up to you! — just that the underlying data indicates their placement is probably higher, possibly significantly so, than the conventional wisdom suggest.

Third, there are two quality of life variables used. One comes directly from the academic literature. Professor David Albouy has conducted some great and fascinating research on metropolitan areas, including both quality of life measures and housing cost differentials (both used in the trilemma work). It is a composite measure that looks at the characteristics of a place for which people are willing to pay more to live in, after controlling for incomes, essentially. The underlying variables include things like weather, distance to the ocean, crime, air quality, bars and restaurants, and arts and culture. The second variable used is household purchasing power, effectively getting at how far one’s dollar goes in a particular place after adjusting for the cost of living. Combined these measures get at nice places to live where you can afford to do so. They are not perfect measures, of course, but do provide fairly reasonable rankings across all 100 metros. The Great Plains/Midwest metros tend to rank in the 30-50 range.

Posted by: Josh Lehner | June 8, 2016

The Housing Trilemma

Every city wants to have a strong local economy, high quality of life and housing affordability for its residents. Unfortunately these three dimensions represent the Housing Trilemma. A city can achieve success on two but not all three at the same time. Underlying all of these tradeoffs are local policies as well.

Inspired by Kim-Mai Cutler and Cardiff Garcia, I set out to try and quantify the Housing Trilemma across the nation’s 100 largest metropolitan areas. It turns out to be very real. Just eight rank among the top half for all three dimensions of the Housing Trilemma. None rank among the Top 20 in all three. Unless you prefer living on the Great Plains, that list of eight metros lacks sizzle*.

Update: All 100 MSAs fall within the Venn diagram below. The metros listed are to show some individual metros that are representative of each part of the diagram. For example, New York City slots in right between San Francisco and Portland. Kansas City is just to the right of Oklahoma City. And so forth. Email me if you would like a version with your metro on it.


The reason these tradeoffs exist is mostly, but not entirely, due to market forces. People want to live in cities with a strong economy and high quality of life. Increased demand for housing leads to higher prices and lower affordability. Nice places to live get their housing costs bid up due to strong demand. The opposite is true as well. Regions with underperforming economies and a lower quality of life do have better affordability.

Below are the Housing Trilemma graphs for Portland, Houston and Youngstown to give a few examples of how each city compares to the nation’s 100 largest metros. Note that the bars represent percentiles, or rankings. This is to make the graph easy to read and visually intuitive. The longer the bar, the better a particular metro ranks on that measure. Click here for an interactive version** where you can pick and choose any metro to examine and/or download the data. Go to the “graph” tab and choose a metro from the dropdown menu. Take it for a test drive. The specifics of each metric are shown at the end of the post.

UPDATE: Some are noting issues with the interactive version. Download in Excel here: HousingTrilemma

Trilemma-PDX Trilemma-HoustonTrilemma-Youngstown

Google Docs Version  or Excel Download: HousingTrilemma

Clearly a few patterns emerge. In particular the popular metropolitan areas stand out, not least because their eroding housing affordability is constantly discussed. What you could call the cool city profile is seen in the Denvers, Portlands and San Franciscos of the world. In a way, they are victims of their own success. Their strong regional economy and high quality of life do come as the cost of lower housing affordability.

Yet even among this group, affordability does vary. Portland is an extreme case with significantly more households cost-burdened and a lower vacancy rate than nearly all other metros in the nation. This impacts renters the most, including younger households and those on fixed incomes.

For these popular metros, more construction is required, but that alone is not enough. Just look at Austin, TX. The region has a very strong economy and high quality of life. Despite leading the nation’s largest metros in new construction, Austin is only able to reach middling affordability. Austin’s home prices, while lower than Portland’s or Seattle’s, are still relatively high and half of all renters are cost-burdened. Increasing construction is able to help with broad, regional affordability, but cannot fully offset the premium required to live in a popular place. In addition to building more homes, targeted programs are also needed to help less fortunate neighbors bear these costs.

The housing trilemma is real. Tradeoffs are inevitable. Here in Portland and the Northwest more broadly, we are fortunate enough to have a good economy and desirable quality of life. We should work to maintain these successes. However, eroding affordability in Portland does not have to be a permanent trend. Increasing construction to match a growing population and strong assistance programs are needed.

Here are the exact measures used.


* Full disclosure: I am from the Great Plains and many of my relatives still live there. However, my wife likes to joke that it’s a great place to be from. These metros usually don’t knock it out of the park on any particular measure, however they are successful and do rank relatively well when compared to the rest of the nation’s largest metros.

** Apologies for the external dataviz link. Having difficulty embedding the work on our site. Let me know if you have any questions.

Older Posts »



Get every new post delivered to your Inbox.

Join 563 other followers